I present a formal model of symmetric n-firm Cournot oligopoly. The set of firms is divided into two subpopulations of imitators and optimizers. Imitators mimic the output decision of the most successful firms of the previous round. Optimizers are myopic best response players to the opponents’ previous output. The dynamics of the decision rules induce a Markov chain. Firms are allowed to make mistakes and deviate from the decision rules with a small probability. Applying stochastic stability analysis I characterize the long-run behavior of the oligopoly. It is known that a homogeneous population of optimizers converges to a Cournot Nash equilibrium. Vega-Redondo (1997) showed that a homogeneous population of imitators converges to the competitive outcome. I find that a heterogeneous population converges to a recurrent set of states where imitators are better off than optimizers.